Using invariant feature descriptors for an efficient image retrieval

In this paper, we study some feature descriptors and detectors based on invariants type and use the most stable and robust keypoints for an efficient object detection. From many points of view local descriptors are relatively different, the way of their extraction and the type of the invariance, the goal of this paper is to examine existing feature detectors and apply the most appropriate and efficient on our Content-based image retrieval engine (CBIR).